Cost-effective and reliable damage detection and health monitoring of critical members are very essential for safe operation of civil infrastructures such as bridges, buildings, power plants, etc. This dissertation addresses several primary software and hardware issues for online structural health monitoring (SHM) techniques for critical members of civil infrastructures using piezoelectric active sensors. The piezoelectric active sensors can provide special features suitable for real-time wireless SHM systems of the critical members, because they are small, light, cheap, and useful as a $built-in$ actuator/sensor system, and show a small power-requirement. At first, research backgrounds and motivations for current studies are described, and previous research works related to the topics of active sensing-based SHM techniques using the piezoelectric sensors are discussed. Wireless SHM systems are also reviewed briefly. Active sensing-based SHM systems composed of several piezoelectric active sensors are presented in conjunction with both electro-mechanical impedance-based method and guided waves-based method to detect incipient damage which may occur on the critical members. The active sensing-based SHM systems are developed using statistical pattern recognition algorithms such as support vector machine (SVM) and outlier analysis (OA) to identify the location, severity, and type of damage on the critical members. The pattern recognition algorithms analyze statistical distributions of the damage sensitive features (such as root-mean-square-deviation (RMSD) for the impedance signatures and wavelet coefficients (WC) for maximum energy mode of the guided waves) derived from the measured data to enhance damage detection and classification processes. The feasibility of the proposed SHM strategies is investigated through a series of example studies using a steel bridge member, an aluminum plate, and a rail specimen. With a current trend of SHM heading towards unobtrusive self-contained sensors, an active sensing node incorporating on-board microprocessor and radio frequency (RF) telemetry is utilized in a sense of tailoring the impedance method to wireless sensing technology. A data compression algorithm using a principal component analysis (PCA) is embedded into the on-board chip of the active sensing node to provide better performance for damage detection and to reduce the power requirement for wireless data transmission. A new electro-mechanical impedance model is proposed, which incorporates the effects of the sensor and bonding defects for simultaneous achievement of both functions of structural damage detection and sensor self-diagnosis. The feasibility of the modified impedance model is verified by a series of parametric studies using a simple example of a PZT-driven spring-mass-damper system with a single degree of freedom. Finally, this dissertation summarizes several issues and future works that can be used as a guideline for future investigation.
본 논문에서는 압전센서(Piezoelectric sensors)의 능동센싱기법(Active sensing techniques)을 이용하여 사회기반시설물의 주요부재의 구조 건전도를 평가하고 무선으로 온라인 모니터링을 수행하는 데 있어 요구되는 소프트웨어 및 하드웨어에 관한 문제에 대하여 연구하였다. 임피던스와 유도초음파를 기반으로 하는 손상검색 기법을 이용하여 사회기반시설물의 주요부재에서 자주 발생하는 균열, 볼트풀림, 부식 등의 손상들을 찾아내는 방법론을 연구하였다. 특히, 웨이블렛 해석(Wavelet Analysis)과 같은 신호처리기법과 Support Vector Machine이나 Outlier Analysis등과 같은 패턴인식기법들을 적용함으로써, 손상의 검색 및 분류 과정에서 정확성 및 효율성을 증대시켰다. 임피던스 기반 손상 검색기법에 대해서는 내장형 마이크로 프로세서와 무선 송신기를 장착한 무선센서노드를 적용함에 있어, 마이크로 프로세서에 데이터 분석 및 손상평가 성능을 향상시키기 위하여 주성분 분석(Principal Component Analysis)을 이용한 데이터 압축 알고리즘을 개발하여 이를 삽입하였다. 이를 통하여, 데이터 분석능력을 향상시킬 뿐만 아니라 전력 소모 또한 줄일 수 있었다. 마지막으로 구조물의 손상 검색뿐만 아니라 센서의 성능 및 부착상태의 자체진단이 가능한 새로운 임피던스 모델식을 제안하였는데, 수치해석 연구를 통해 그 유효성이 검증되었다. 본 연구에서 제안된 기술들은 향후 실제 사회기반시설물의 주요 부재의 건전도 평가와 경제적이고 신뢰성 있는 유지관리에 효과적으로 적용될 것으로 예상된다.
